A Distributed Collaborative Adaptive Sensing System: A Feasibility Plan for Korea. Sanghun Lim Colorado State University Dec.

Transcription

1 A Distributed Collaborative Adaptive Sensing System: A Feasibility Plan for Korea Sanghun Lim Colorado State University Dec

2 Outline q The DCAS concept q X-band Radar Network and severe storms q QPE at X-band Radar Network q Deployment of Urban Flooding Monitoring System: CASA-NIED Partnership q Summary Dec ,

3 Overview of Networked Sensing DCAS Radar Network Dec ,

4 There is insufficient knowledge about what is actually happening (or is likely to happen) at the Earth s surface where people live. [NRC, 1998] 10,000 ft 3.05 km 1 km 2 km 4 km snow wind tornado 5.4 km gap Horz. Scale: 1 = 50 km Vert. Scale: 1 -=- 2 km RANGE (km) Distributed short-range radar network can improve the coverage of lower level atmosphere. Dec ,

5 CASA Testbed q CASA developed a network centric sensing paradigm DCAS: Distributed Collaborative Adaptive Sensing IP1 test bed: the first DCAS radar network Southwestern OK ~ 7000 km 2 4 radar nodes X-band Dual-polarization Dec ,

6 Where and when user needs are greatest 1. Coordinated volume scan by radar network 2. Real time data mining to identify important features to users Now Cast 4. Dissemination of Data Products to Users 3. Optimization for radar control : task generation user preferences, radar capabilities, meteorological phenomena Dec ,

7 IP1 System Architecture q IP1 radar has the capability of rapid adaptive scans q Dynamic & Adaptive sensing in a closed-loop control q Optimize the system resource for weather and users need. Adaptive rapid sector scans ensure good temporal resolution and multiple network view. Closed-loop Control The network has demonstrated a great potential to detect and track tornadoes. Dec ,

8 IP1 System Architecture The Anadarko Tornado and qdynamic Damaging & Adaptive Wind Event Sensing: optimize the system resource May 13, against 2009 weather and users need. Linking System Operation and Response Adaptive rapid sector scans ensure good temporal resolution and multiple network view. Closed-loop Control Dec ,

9 Anadarko Tornado and Damaging Winds Event May 13, 2009 EF2 Tornado ~9:22 9:40PM Prolonged Damaging Winds 100mph+ 3 injuries $43 million+ in property damage CASA Tornado Warning 9:21 NWS Tornado Warning 9:24 Dec ,

10 Real-time Doppler Wind q Real-time 2D Doppler Wind Dec ,

11 QPE Aspects of Networked Sensing How good is X-band radar QPE? Dec ,

12 K dp based Radar QPE q At X-band, attenuation and differential attenuation can be large and must be compensated. q K dp has better response to rainfall at X-band. q R- K dp based radar QPE is especially appealing for X- band radar network - üindependent of radar calibrations in the network; üavoids the uncertainty due to attenuation correction; A scaled version of KOUN s rainfall estimation is implemented (based on local measured DSDs). R = K dp mm/hr R = K dp mm/hr : KOUN : IP1 Dec ,

13 Adaptive Estimation of Kdp q K dp, as the derivative of F dp, can be very noisy. qadaptive estimation: Derivatives over large distances in light rain region Derivatives over short distances in heavy rain to keep up the scale q An adaptive algorithm is developed using the vectorized F dp profiles. Working over wrapped phase Dec ,

14 Network Composition q The data quality of K dp differs on different propagation path, mainly depending on the cross-beam gradients. merging Dec ,

15 Validation Study Dec ,

16 IP1 QPE Validation Study q Gauge comparison was investigated to evaluate the QPE system q USDA ARS Micronet A rain gauge network located at the center of the IP1 test bed Source: Little Washita Watershed size: 611 km 2 Mean annual precipitation: 760 mm Gauge network: 20 tip-bucket stations Dec ,

17 IP1 QPE Evaluation q Is R(K dp ) feasible for high-resolution rainfall mapping at X-band, on an operational perspective? q Metrics: Normalized bias: Normalized error: q 29 Events: 2007: 6 events 2008: 8 events < e > 2009: 15 events (up to the end of June) q Overall performance N NSE = = R R R R R - R R G G - R G G Instantaneous Rain Rate FB: 4.16% NSE: 49.00% Hourly Rain Accumulation FB: 4.26% NSE: 22.76% Dec ,

18 Sample Gauge Comparison q May 07, 2007 Flood warning issued over the Micronet area Dec ,

19 NIED s X-NET QPE Study q A composite R-K dp based estimation was implemented in NIED s (National Research Institute for Earth Science and Disaster Prevention, Japan) X-Net test bed. q X-Net Ground Validation Three events: 2 straitform, 1 typhoon R = 19.63K K dp dp > 0.3 o mm/hr / km and Z h > 35dBZ R = Z h mm/hr Also based on locally measured DSDs Dec ,

20 Operational Feasibility q Peer multi-year operational evaluation at S-band: Compared to KOUN s operational evaluation 24 rain events (4/2002-7/2003) Ryzhkov, A.V., S.E. Giangrande, and T.J. Schuur, 2005: Rainfall Estimation with a Polarimetric Prototype of WSR- 88D. J. Appl. Meteor., 44, Hourly Rainfall Accumulation Radar System or Network Total Events Analyzed Normalized Bias (%) NSE (%) IP MP-X Dec ,

21 Discussion: Network Aspect q The diversity on IP1 observations can improve the network coverage and QPE accuracy. q Example case: April 27, 2009 (UTC) Hourly Rainfall Estimation Radar <e> (mm) <e> N (%) NSE (%) Chickasha Cyril Lawton Rush Springs Network Composition Dec ,

22 Discussion qother factors likely contribute to the improvement: Resolution, short radar range Lower beam height Better K dp estimation Floods Warning Hydrological Models High Resolution Radar QPE Accuracy Ground Validation Temporal Resolution Observation Spatial Resolution Low Level Adaptive Collaborative Short Range DCAS Radar Network Dec ,

23 Urban Flooding Monitoring System: CASA-NIED Partnership Dec ,

24 X-NET Topology MKA Radars networked in X-NET EBN, KSR, MKA (NIED) ABK (CRIEPI) KFU (Yamanashi Univ.) BNK (Chuo Univ.) YKS (NDA) HNY (JWA) YKH, STM (MLIT) Research Polarimetric Operational Polarimetric (plan) Research Doppler Operational Doppler (C-band) KFU EBN HNY BNK STM Metropolitan Area YKH YKS KSR ABK Synthetic Doppler radar analysis area km Synthetic Doppler radar analysis area Dec ,

25 Specifications of NIED polarimetric radars Code name MP-X (EBN) X-POL(MKA) X-DOP(KSR) Photo Frequency GHz GHz GHz Antenna Type 2.13 mf Parabola 2.2 mf Parabola 2.2 mf Parabola Antenna Gain 41.6 db 44.2 db (with rd) 44 db Beam Width (H),1.09 (V) 1.1 Transmitter Tube Magnetron 50kW Magnetron 40kW Klystron 50kW Pulse Length 0.5 ms 0.5/1.0 ms 0.5/1.0/2.0 ms PRF 1800 Hz 1800 Hz 1800 Hz Min. Detectable Signal -111 dbm -114 dbm -115 dbm Observation Range 80 km 80 km 80 km Outputs T, Z, V, W, Z DR, r hv, T, Z, V, W, Z DR, r hv, T, Z, V, W, Z DR, r hv, F DP, K DP F DP, K DP F DP, K DP Dec ,

26 Operational Networks in Japan q Radar networks are being deployed the by Ministry of Land, Infrastructure, Transport and Tourism (MLIT). q X-NET will prepare the algorithms for the operational MLIT networks. Major Japan Metropolitan Areas Toyama, Kanazawa Tokyo population: 34 million area:13,754km 2 Fukuoka Hiroshima Osaka Nagoya Tokyo Osaka population: 9 million area:6,380km ~ 2011~ Nagoya population: 19 million area:11,169km 2 Dec ,

27 EARLY WARNING SYSTEM EXAMPLE MEDELLIN (COLOMBIA) Dec ,

28 EARLY WARNING SYSTEM EXAMPLE MEDELLIN (COLOMBIA) Main hydrometeorological sensing resource Complex terrain deployment Dec ,

29 Possible DCAS applications at Korea Dec ,

30 Seoul Metropolitan area 35 km Dec ,

31 Gangwon Province and Lake Soyang area Dec ,

32 Jirisan National Park area Dec ,

33 Thank you! Dec ,